- Email: [email protected]
The pathways of dynamic recrystallization in all-metal hip joints
Wear 259 (2005) 887–897
The pathways of dynamic recrystallization in all-metal hip joints夽 R. B¨uscher ∗ , A. Fischer Department of Materials Science II, University of Duisburg-Essen, Lotharstr. 1, 47057 Duisburg, Germany Received 31 August 2004; received in revised form 12 January 2005; accepted 1 February 2005 Available online 10 May 2005
Abstract In the present study metal-on-metal (MoM) McKee-Farrar prostheses of the first generation were investigated by means of transmission electron microscopy (TEM). It was possible to observe the worn regions using a novel taper section preparation technique without producing any artefacts. Thereby, it could be shown that the wear in vivo leads to a reduction in grain size by a factor of up to 20,000. This is achieved by recrystallization via two pathways which act simultaneously within the subsurface regions. One is dominated by the metallurgical characteristics of the material and follows the gradient of the friction induced shear strains from the bulk towards the surface. The other one is merely acting within the tribological contact volume directly at the surface. Both mechanisms lead to a significant increase in strength. © 2005 Elsevier B.V. All rights reserved. Keywords: Recrystallization; Nanocrystals; Wear; Artificial hip joints
1. Introduction With approximately 750,000 surgeries per year the total or surface replacement of the natural hip joint is one of the most common surgeries in orthopaedics [1]. Artificial hip joints have been used since 1938 and are nowadays implanted with a high chance of success, e.g. 97% in 2000 [2]. However, the postoperative rate of revisions is still high with 50% in 1996 [3]. Although many different effects like individual medical predisposition or life style and to a much lesser extent clinical mistakes influence the life time of artificial hip joints, the most dominant reason for early failure is still induced by cellular foreign body reactions resulting from the emission of wear particles [4]. Therefore, minimizing wear in the articulating contact areas is still the major goal to attain orthopaedic implants with a longer lifetime. The tribological demands to achieve clinical lifetimes beyond 15 years are currently fulfilled by four material combinations: metal-on-polymer, ceramic-on-polymer, ceramic夽 To be presented at, Wear of Materials Conference 2005, San Diego, CA, USA. ∗ Corresponding author. Tel.: +49 203 379 1266; fax: +49 203 379 4374. E-mail address: [email protected] (R. B¨uscher).
0043-1648/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2005.02.036
on-ceramic and metal-on-metal. The choice of material does not only depend on the friction and wear properties but is also strongly influenced by the expected lifetime and activity – thus the age – of the patient, the experience of the surgeon and finally also the financial and legal situation of the clinic. In the US, as the world’s largest market with about 270,000 total hip replacements each year [2], metal-on-polymer couples are still referred to as the “golden standard”. However, a comparison of the wear rates of the different pairings shows that the classic Charnley prosthesis (metal-on-polymer) generally leads to higher clinical wear rates (<500 m/a) when compared to hard-hard couples such as ceramic-on-ceramic and metal-on-metal joints. The latter are believed to attain very low wear rates of 1–5 m/a after an intense running-in wear in the order of 35 m/a. The fact that some of the joints from the first generation of MoM hip joints (e.g. McKeeFarrar prostheses) lasted for more than 20 years, led to a promising renaissance of all-metal pairings. Nevertheless, the metallurgical mechanisms, which contribute to high runin wear as well as to low steady state wear rates are yet to be understood. Therefore, the present study will focus on the metallurgical investigation of the microstructural changes, which lead to the acting in vivo wear mechanisms of MoM hip prostheses.
888
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
2. Materials and methods
Table 2 Clinical data of the retrieved and studied McKee-Farrar couples (from [5])
Since the number of retrievals is always limited and – as to the nature of human life style – in vivo wear process takes place under distinctly unidentified parameters (starting conditions, loading sequences, duration, and joint motion characteristics, etc.) specimens from a conventional disc-on-pin (DoP) wear test were included for comparison in order to investigate the impact of mechanically dominated wear mechanisms on the subsurface microstructure. Thus all microstructural investigations were performed on laboratory wear specimens and on retrievals.
Element
Couple A
Couple B
Couple C
Alloy trade name Manufacturer
Coballoy Dow, UK
Head diameter Age of patient at revision Duration in vivo
35 mm 80 a
Vinertia Howmedica, USA 40 mm 69.4 a
Vinertia Howmedica, USA 35 mm 85.3 a
7.5 a
17.8 a
19 a
2.1. Laboratory wear specimens Details about the testing equipment and conditions can be found in Table 1. A fixed cylindrical pin with a diameter of 6 mm and a spherical tip (radius: 16 mm) runs against a rotating disc with a diameter of 30 mm. Both are emerged in distilled water. The radius of the as generated wear track was 9 mm and the frequency of testing was 1.77 Hz. A normal load of 5 N was applied resulting in a theoretical maximum Hertzian contact pressure of 370 MPa. Prior to testing, the contacting surfaces were mechanically polished to a Ra value of 0.02 m. The total sliding distance was 10.4 km corresponding to 30 h testing period. This would be roughly 10 weeks in vivo. It should be made clear that the focus of this additional laboratory study was not to simulate body conditions but to solely investigate and understand microstructural changes under tribologically induced shear stresses. This should be done under the action of mechanically dominated wear mechanisms, which are similar to those in vivo. Certainly the influence of chemically dominated wear mechanisms is not regarded. But with respect to the subsurface shear stresses they might be neglected in a first approach. Incorporating the chemistry will be a substantial part of the authors’ future work after the metallurgy has been fully understood. 2.2. Retrievals (in vivo) Three metal-on-metal (MoM) couplings were chosen from a set of 42 retrieved McKee-Farrar prostheses on the basis of Table 1 In vivo conditions vs. laboratory test conditions
typical wear appearance. The 42 retrievals were in situ for 13.6 years at average (range 1, 3 . . . 22) and none of them was removed for the reason of excessive wear. At the time of removal the prosthesis were carefully rinsed to remove blood and subsequently sterilized and packed. Caution was taken not to touch or damage the bearing surfaces. The entire collection of 42 retrievals has been obtained from a single surgeon and intensively studied by the authors with respect to wear appearances, wear mechanisms and their sequence of acting [5–9]. For this contribution three pairs were solely selected on the basis of macroscopic wear appearances. There was no effort undertaken to control for additional parameters like e.g. manufacturer or time in situ. All three implants showed wear marks in the form of fine scratches or grooves as well as tribochemical reaction layers [9], but no major damage of the articulating surfaces was observed by the unaided eye. Details about brand name, manufacturer and clinical data of those three prostheses are summarized in Table 2. 2.3. Materials The laboratory testing material was a forged low carbon CoCrMo-alloy (Endocast SL, gb Implantattechnologie, Essen, Germany), with a chemical composition similar to that of the cast McKee-Farrar prostheses (Table 3) The major differences lie within the grain size of 600 m for the cast endoprotheses and of 37 m for the forged alloy. In addition the latter does not contain as many carbides. However, neither Endocast SL nor the McKee-Farrar alloys belong to the Table 3 Chemical composition (in wt.%) of the studied low carbon wrought alloy CoCr29Mo6 (Endocast SL) relative to the McKee-Farrar alloys (Coballoy, Vinertia) and ISO 5832/12 [56]
Parameter
Cup-on-head
Disc-on-pin
Element
Endocast SL
Coballoy
Vinertia
ISO 5832/12
Contact geometry
27.5 5.37 0.008 0.17 <0.002 0.26 0.08 0.01 0.42 0.53
26–30 5–7 <0.35 <0.25
<1 2.9
<1 2.3
<0.75 <1
<1
<1
Synovial fluid
Cr Mo C N S Fe Ni W Si Mn Al Co
27.4 5.4
Medium
Flat versus convex (radius: 16 mm) 5N 370 MPa 0.1 m/s (1.77 Hz) 22 ◦ C Unidirectional 30 h (2.8 km sliding distance) Distilled water
26.4 4.5
Normal force Hertzian pressure Relative velocity Temperature Sliding motion Duration of loading
Concave versus convex <2000 Na <20 MPab <0.2 m/sa (1 Hz) 37 ◦ C Multidirectional 7.5 a
<1 <1
<1 Balance
<1 Balance
a b
Unsworth [54]. Hodge et al. [55].
Balance
Balance
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
so-called high carbon (HC) cast or forged alloys, which had been investigated earlier [8]. Certainly, the microstructure, morphology and volume fraction of carbides strongly influence the wear rate. Thus, this work will not compare wear rates and try to generalize the results for the various alloy types but will address the basic microstructural changes of low carbon cobalt base alloys under shear strains induced by sliding wear. 2.4. Metallography, microscopy, and hardness testing
889
specimens were further thinned from both sides by means of dimple grinding (Model 656, Gatan GmbH, Munich, Germany) and ion milling (PIPS 691, Gatan, Munich, Germany) down to 100 nm. TEM investigations were performed with an accelerating voltage of 120 keV. In a recent study with worn high-nitrogen steel samples it has been demonstrated that this technique does not introduce artefacts [10]. Since the physiological gait cycle creates a multidirectional movement on the surface of the hip joint, the orientation of the in vivo taper section can not be chosen according to a certain direction of relative motion. However, the DoP wear test is unidirectional and, therefore, this allows for the defined orientation of the samples during preparation. But, due to the fact that the wear track is narrow and circular the chosen TEM preparation technique allows only a preparation of a transversal taper section. Thus, the figures will show the microstructure underneath the wear track perpendicular to the sliding direction. In order to evaluate the strain induced hardening of the subsurface microhardness measurements were carried out by means of a Vickers indenter and a load of 10 p (=98 mN, HV0.01). The hardness gradient was measured at 40 positions within a distance of up to 40 m below the surface.
In order to gain information about the hypothetical changes of the microstructure underneath the articulating surfaces, in vivo and DoP samples were investigated in taper sections (=90◦ cross section) using a light microscope (LiMi) as well as a transmission electron microscope (EM 400, Phillips, Eindhoven, The Netherlands). For LiMi taper sections from heads and cups as well as from disc’s and pins, samples were prepared by wet cutting, grinding, polishing and finally chemical etching for 45 s in 50 ml H2 O + 50 ml HCl + 4 g K2 S2 O5 . For TEM a conventional preparation technique was employed using two parallel cuts of the worn areas with a thickness of 500 m, which were glued together with a twocomponent adhesive. Fig. 1 shows a medial and parallel cut of the prostheses with the sample regions indicated with 1 (cup sample) and 2 (head sample). To minimize the gap between the samples, a convex sample of the head was glued together with a concave counterpart from the joint cup. The preparation of the DoP samples was much easier done by glueing together two pieces of one disc. These prepared samples were then fixed in a brass cylinder (diameter: 3 mm) using a slotted pipe (diameter: 2.5 mm) as well as a suitable adhesive (Epoxy G1, Gatan, Munich). A heat treatment at room temperature for 30 min and at 150 ◦ C for 2 h resulted in a sufficient bonding strength of these composite structures. Afterwards it was cut into 400 m thin slices by slow speed wet cutting. After grinding to a thickness of 100 m, the
At a certain magnification (Fig. 2) a distinct degree of plastic deformation can be observed below the worn track in the direction of the tribologically introduced shear stresses. According to the LiMi observations, TEM reveals different appearances of plastic deformation with respect to the distance to the worn surfaces. At a depth of about 30 m from the worn surface stacking faults and -martensite needles prevail (Fig. 3). Electron diffraction patterns of the needles verify -martensite, which has a hexagonal closed packed
Fig. 1. In vivo; cross-sectional overview of a cut McKee-Farrar prosthesis.
Fig. 2. Light microscopic cross-section of the wear track. The arrow indicates the sliding direction of the pin. NC = nanocrystalline zone.
3. Results 3.1. Laboratory wear specimens
890
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
Fig. 3. Coexistence of stacking faults and -martensite needles about 30 m below the worn surface (laboratory sample).
Fig. 5. Nanocrystalline microstructure at the disc’s surface (laboratory test).
(hcp) lattice structure and form strictly on the discrete (1 1 1) sliding planes [11]. These lattice defects react with each other under higher shear strain (= closer to the surface) forming an arrangements of, e.g. rhombic cells (Fig. 4). Between depths of about 10 and 1 m no clear microstructure can be identified. Moreover, the metal appears to be in some state of severe plastic deformation. Nevertheless, closer towards the surface at depth of less than about 1 m again a distinct microstructure of very fine granular crystals in the nm range occur (Fig. 5). The microhardness profile (Fig. 6) shows the cold working of the CoCrMo solid solution by the observed lattice defects. Even though the measured values scatter, the tendency is unequivocal. Starting from the microhardness of the unworn material of 450 HV0.01 the hardness increases at about a depth of 30 m up to 660 HV0.01 at 3 m. Closer to the surface the hardness might even be higher, but this cannot be
measured by such microhardness techniques. The measurements adjacent to the wear track prove that the influence from mechanical polishing can be neglected since no increase in hardness was observed within the same region. Therefore, the changes in hardness should be a result of the wear-induced changes of the subsurface microstructure. The coefficient of friction in the steady state wear regime was µ = 0.3.
Fig. 4. Rhombic cells generated by intersecting -martensite needles about 4 m below the worn surface (laboratory test).
3.2. Retrievals LiMi observation reveals the coarse grain structure of the cast material but no significant changes of the microstructure in the vicinity of the contact area as depicted in Fig. 7. Nevertheless, the results of the TEM investigation show a different picture, which appears similar to that of the laboratory specimens. The microstructure below the articulating surface has changed significantly because of tribological stresses. This was observed on both sliding partners of all investigated pairs. Again in a certain depth from the surface, e.g. twins and martensite needles form on the sliding planes of the CoCrMo solid solution (Fig. 8 ). In other areas stacking faults appear as well. Again these lattice defects react with each other generating triangular or rhombic cells. Closer to the surface the grains become nanocrystalline (Fig. 9). The microhardness profile of this retrieval also shows (Fig. 10) cold working but the measured values scatter much more distinct than for the laboratory specimens. Already within the bulk material the values scatter between 310 and 360 HV0.01. This can be simply attributed to the effects of microsegregation within cast microstructures, which show the characteristic dendritic structure within large grains (Fig. 7). Despite the scatter again the work hardening tendency towards the surface is apparent. Beginning with the average microhardness of the unworn material of about 340 HV0.01 the hardness increases at about a depth of 15–20 m up to 390 HV0.01 at 3 m from the worn surface.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
891
Fig. 6. Depth profile of microhardness HV0.01 (laboratory test).
Due to the distinct scatter there is no use in any attempt to quantify the work hardening effect, but in comparison to the laboratory specimens less volume is affected and less work hardening is noticed in the region below 2 m from the worn surface. In addition the in vivo gradient is flatter. Nevertheless, the work hardening is induced by tribologically introduced shear strains leading to lattice defects and phase transformation starting from about 30 m below the surface.
4. Discussion MoM hip joints are self-mating tribological systems in which sliding wear prevails. Following the literature it has been known for decades that because of their high ductility fcc metals – like CoCrMo alloys – are prone to wear by adhesion [12]. However, neither on the worn surfaces of the joint nor on the surfaces of the laboratory samples any signs
Fig. 7. Light microscopical cross-section of McKee-Farrar hip cup. Arrow points on the articulating surface.
of adhesion were present. As indicated by the authors tribochemical reactions are present on the surfaces and generate layers of denaturated organic material. These layers separate the surfaces and thereby hinder direct metal-on-metal contacts [8,9]. However, this effect is supposedly happening on discrete spots on the surfaces and cannot be responsible for the total absence of adhesion. TEM investigations revealed in addition that the uppermost surface consists of a nanocrystalline microstructure of fcc and strain induced hcp CoCrMo solid solutions [11,13]. These fine grains bring about surface fatigue in a nm-range generating the well known globular and needle like wear debris in the nm range, which also can act as solid lubricant and separates the contacting surfaces. Beside the hypothesised – but not verified – extreme cold working effect of such a fine grain and lattice defect rich zone this change in microstructure has certainly a big impact on the wear mech-
Fig. 8. Formation of rhombic and triangular cells by reaction of lattice defects in a distance of about 5 m from the worn surfaces (McKee-Farrar, retrieval B, cup).
892
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
Fig. 9. Nanocrystalline surface of McKee-Farrar Prostheses (retrieval A, cup).
anisms as has been shown by DoP laboratory tests [11]. Due to the gradient of properties towards the worn surface no instabilities occur which would represent eventual sites for crack initiation. Moreover, as a consequence of the low stacking fault energy there is a total absence of dislocation cells below the worn surfaces. Thus tribologically induced fatigue crack initiation and propagation is shifted towards the top nanocrystalline surface generating just nm small delaminating filaments. Such small filaments are easily crumbled into nanoscopic single particles, as they have been observed in several in vitro and in vivo studies [14–19]. Thus, owing to the low stacking fault energy of CoCrMo alloys together with the bio-tribochemical reactions on the worn surfaces MoM hip joints achieve their low clinical wear rates. A very similar development of lattice defects and metallurgical phase transformation was observed for both tribosystems below the contact areas. However, the depths in which
the different effects take place differ, which mainly can be attributed to different loading conditions. Nevertheless, the thickness of the nanocrystalline zone as well as the grain size within it did not differ significantly between the DoP and the in vivo samples. Fig. 11 shows the distribution of grain sizes. Although the mean grain size below the discs surface (37 nm) is slightly higher than in vivo (32 nm) this should not be overestimated. The thickness of the nanocrystalline layer in both tribological systems is well below 1 m; with being slightly thicker on the laboratory sample (800 nm) than in vivo (550 nm). Now the question remains why materials with such a low stacking fault energy – hindering cross slip and climbing – recrystallize at room temperature from 600 m (cast) and 37 m (forged), respectively to about 30 nm. In addition it is not clear why the final grain size in the worn surfaces of both tribosystems are similar, even though the loading conditions and as a consequence the hardness gradients are different. Thus on the basis of the analyses as well as on existing literature at first the strain hardening and at second the recrystallization should be discussed. Due to the fact, that the metallurgical findings of the in vivo and lab specimens are similar both will be treated together. Different observations are mainly due to the different loading conditions and movement characteristics, respectively. Larger contact pressures and, therefore, supposedly higher friction shear forces as well as the unidirectional relative motion lead to a more distinct straining of the DoP subsurface microstructure compared to the multidirectional character of the in vivo tribosystem with unknown coefficient of friction. 4.1. Microstructure and hardness The strain induced hardening of the subsurface is achieved by different microstructural effects, which act simultaneously. Generally, blocking and subsequent accumulation of dislocations lead to an increase in hardness. Blocking can
Fig. 10. Depth profile of microhardness HV0.01 (McKee-Farrar retrieval A, head).
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
893
Fig. 11. Comparison of the grain size distribution in the nanocrystalline zone of in vivo and laboratory specimens.
be obtained from plastic deformation by either increasing the grain boundary density or creating other lattice defects which also hinder the movement of dislocations. In particular, stacking faults and their intersections are known to distinctly increase the strength by piling up dislocations [20]. Merely at elevated temperatures these intersections may be passed by dislocations to achieve a relaxation of the lattice [21]. However, this form of work hardening is even limited at room temperature since the stacking fault density is increased with increased work hardening which in turn hinders the formation of new stacking faults [22]. In addition to stacking faults a strain induced phase transformation into hcp needles further contribute to this hardening effect. It should be mentioned that the hardening effect of strain induced -martensite is similar to that of stacking faults and differs from that of ␣ -martensite in steels. In addition to these grain boundaries must be regarded as well. The quantitative increase in hardness which can be achieved by grain boundaries was described by Hall and Petch in the early 50’s. Their relationship predicts an increasing hardness with a decreasing grain size (Hardness ∼ 1/diameter0.5 ) [23,24]. This is confirmed for cobalt base alloys when the hardness of fine grained wrought alloys (400–500 HV0.05) is compared to the hardness of cast alloys (300–350 HV0,05) having much coarser grains. However, nanocrystalline metals do not always follow the HallPetch relationship. Umemoto [25] showed that the hardness of nanocrystalline Al, Ti, Ni and Fe is even higher than it was calculated referring to Hall-Petch. In contrast any extreme reduction in grain size might also lead to a softening of the material. This effect, which is often referred to as inverse Hall-Petch relation, is valid for grains which are smaller than dislocation loops and hence, no hardening will take place [26–28]. Fine grains also are able to rotate and slide relative to each other thereby leading to a plastic deformation which is not brought about by the generation and movement of dislocations. Thus hardness might not alter or even decrease [29].
For the cobalt base alloys investigated here lattice defects were still visible within the nanocrystals. Thus it is supposed that the inverse Hall-Petch relation does not apply and the hardness is enhanced towards the surface. This is also supported by the fact that by means of molecular modelling it has been shown that the inverse Hall-Petch relation applies at a grain size below 5–8 nm [30]. Referring to the tribological behaviour of a nanocrystalline surface, both, the increase in hardness as well as the increase in plasticity due to grain rotation and sliding should be beneficial. The latter may allow some small foreign body particles to be entrapped within the surface rather than initiating abrasion, while a high hardness on the other hand reduces the effect of three body abrasion by increasing the resistance against surface fatigue by indentation. 4.2. Recrystallization A comparison of the bulk and surface grain sizes reveals their significant reduction. For the studied McKee-Farrar joints, the mean grain size of the as-cast cobalt base alloy is in the order of 600 m. In contrast, the crystals at the surface are 30 nm in size. Hence, a recrystallization took place which reduced the original grain size by a factor of approximately 20,000 times within a zone thinner than 100 m. For the DoP samples this factor is lower due to finer initial grains of the wrought alloy. Nevertheless, even for the wrought metal the original grain size is reduced by more than 1000 times. Due to the fact that all materials with low stacking fault energies do not recrystallize as easy as e.g. Cu- or Al-alloys by the generation of sub grain boundaries or cross slip and climbing of dislocations the distinct decrease in grain size is worth being investigated. Stacking faults do not have this capability. Thus, CoCrMo alloys require a very high degree of deformation and a very high temperature to be statically recrystallized. Now the question remains, whether this type of recrystallization –
894
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
which maybe is a dynamic one - is either related to thermal or mechanical effects or both. The characteristics of the tribological system in vivo might help to answer this question. Due to the convex (head) – concave (cup) pairing of the artificial hip joint, the conformity of surfaces is very high which generally results in large contact areas, low local contact pressures and finally relatively low contact temperatures. In addition, the low carbon alloys provide fairly smooth surfaces, which reduce the probability of high medium or flash temperatures. Therefore, the contact temperatures should not be in a range above 1100 ◦ C where a thermally induced recrystallization might occur. Even with HC-CoCrMo alloys, where the contact area may be much smaller due to protruding carbides, a mean temperature increase of only T = 60 K was calculated using an approach by Kuhlmann-Wilsdorf [8]. Thus, mainly mechanically dominated effects must lead to the significant reduction in grain size towards the surface. All MoM joints predominantly undergo mixed lubrication. Hence, the metallic contact between metallic asperities cannot be ruled out. As it has been shown by Koinkar [31], even very low contact pressures can lead to plastic deformations in ductile fcc metals, i.e. that the “shakedown criteria” which predicts only elastic deformation for small pressures may not be valid on a micro- or nano-scale [32]. Thus, the plastic deformation, which is introduced into the subsurface of hip and cup or disc and pin, is responsible for the recrystallization. Such deformed zones are generated by the Hertzian contact stress field directly under the worn surfaces which brings about a constant sequence of static and cyclic strains as described by several authors for materials with high stacking fault energies like Cu, Al, and plain steel [33–35] or at high temperatures [36]. Depending on strain, strain rate and temperature these shear stresses bring about recrystallization via cross slip and climbing of dislocations generating subgrains, which immediately start to rotate or slide against each other. Due to the fact, that this new formation of crystals take space during the deformation process and not afterwards by additional heating this is called dynamic recrystallization. These effects have been also been found with other materials during machining or high velocity impact leading to so called shear bands, in which one finds a nanocrystalline microstructure [37–39]. Now even though the ambient temperature might not be high, the local temperatures within these shear bands during machining are significantly increased [40], which facilitates non-discrete dislocation movement. Under tribological stresses similar recrystallization processes appear even though neither the strain rates nor the temperatures are that high [41]. Thus at least again for Cu- and Al-alloys the observed recrystallization can be explained by the subsurface formation of cell walls, which are known from fatigue loading and are generated by cross slip and climbing as well. If well-defined dislocation cells are formed recrystallization is probable under further plastic deformation, since a high defect density in the cell walls may allow cell rotation and the formation of new grain boundaries [42,43]. The rotation becomes possible even without any further incorporation of
small and wide angle grain boundaries because the diffusion rate of defects and atoms at the cell walls is increased by a factor of 107 –108 when compared to the bulk material [44]. Still CoCrMo alloys have a very low stacking fault energy (SFE) and the ambient and frictional temperatures are fairly low. Thus the mentioned mechanisms of dynamic recrystallization are not valid here. Nevertheless some kind of recrystallization must occur to gain 50 nm grains from 600 m ones. From wear and fatigue laboratory investigations on low SFE materials it is known that cyclic shear stresses bring about localized plastic deformation along discrete sliding planes which generates dislocations, twins, stacking faults and subsequently -martensite [45,46,11]. Under the in vivo tribological stresses the subsurface shear strains activate several discrete sliding planes. On these sliding planes, the generated stacking faults and -martensite needles, which neither cross slip nor climb, react with each other and form the depicted triangular or rhombic cells with a size ranging from 100 to 500 nm (Fig. 8). Certainly these cells block any further sliding of the incorporated lattice defects. Now, due to the steady wear rate any subsurface volume element approaches the worn surfaces and, therefore, undergoes higher cyclic shear stresses. By further increasing the degree of plastic deformation towards the surface, these cells cannot move. Thus, they are sheared which leads to a relative displacement “s” of such lattice defects as highlighted in Fig. 12. If by further shear strains the displacement exceeds a certain value one may speak of a new grain boundary resulting in a highly textured microstructure with smaller grain sizes. This recrystallization mechanism has been theoretically described earlier by Zhang et al. [47]. With the retrieved McKee-Farrar prostheses, this has been predominantly observed in a depth of less than 5 m from the worn surface (Fig. 8), while for the
Fig. 12. Formation of rhombic substructures: rhombic cell C is shifted by the displacement s relatively to A and B along the fcc sliding planes. If s exceeds certain values, a planar mismatch between the cell B and C divides the former macroscopic grain into nanoscopic subgrains.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
DoP specimens this happens in a distance up to 10 m. In this area this type of dynamic recrystallization generated by tribologically induced shear stresses from the moving Hertzian contact stress field is completed. Thus the final grain size is governed predominantly by the size of the former cells. Nevertheless closer to the surface a zone with no distinctly contrasted microstructure is visible. This may be attributed to the high density of sheared cells, which cannot become smaller anymore and might, therefore, start to rotate – similar to subgrains – and subsequently destroy the texture. But at this time it is just a hypothesis. In contrast to this transitional region, the size of the nanograins directly at the surface is smaller than the size of the cells and might be of different reason. Again the nm range of grains size implies some kind of recrystallization mechanism with a similar appearances as it has been found in Cu and stainless steels of various SFEs and were described as “mechanically mixed” and “material transfer layers” [32]. As a matter of fact it could have been shown that a transfer of material does takes place in all-metal pairings which is predominantly directed from the softer to the harder metal [48,49]. In the case of the artificial hip joints such a proof is hard to give since the pairings are self-mating, that means the contacting metals have the same hardness and the same chemical composition. But if this is true the nanocrystalline layer is a mixture of base material, interfacial medium and environment. Thus at least the chemistry at the grain boundaries should differ from that of the grains itself. In addition some of the grains could be of oxidic or organic nature and stem from incorporated remains of some tribochemical reactions at the surfaces. But the verification with such high resolutions is difficult and could not be brought about up to now. Nevertheless, this type of material transfer by mechanical mixing on a nm scale is quite likely at such surfaces. Molecular dynamics showed that contacting surfaces of ductile metals behave like viscous fluids under shearing [50–52]. Hence, the contacting surfaces are exposed to a permanent plastic deformation and rupture in the micro- and nano-scale, by which the nanocrystals are formed by rotations of accidentally generated clusters of atoms within contacting crystals. Thus, the mean grain size within this layer depends only on the flow characteristics of the materials within such a contact. According to Hellstern et al. [29] there is a minimum grain size for each metal, for which a further reduction becomes unlikely due to the need of very high local stresses. In the wear test as well as in vivo, no grains markedly smaller than 10 nm have been observed from which one can assume that this may be the range of minimum grain size for cobalt base alloys to be generated by tribologically induced shear strains. Nevertheless, smaller grains may be tribologically formed without being detected in this work. Due to the fact that in both tribosystems the CoCrMo alloys undergo the same strain range starting from the bulk one can understand why the resulting grains sizes are similar, even though the strain gradients are distinctly different. In addition the results of the TEM essentially indicated that
895
two separate but simultaneously acting mechanisms achieve the strain induced recrystallization in all-metal hip joints.
5. Conclusions and outlook By means of transmission electron microscopy the subsurface microstructures of metal-on-metal McKee-Farrar prostheses were investigated. The results showed that cobalt base alloys are able to adjust their microstructure to the applied tribological loads by a gradual decrease of the grain size. The dynamic recrystallization is achieved by two mechanisms which are supposed to act simultaneously. One is acting at the top surface – by rotating clusters of atoms – within a “mechanically mixed” zone and the second is acting within the bulk – by shearing of cells generated by stacking faults and -martensite needles - moving towards the surface with the increasing strain gradient. Since laboratory tests with wrought alloys revealed a similar tribologically induced microstructure, the wear mechanisms during the steady state wear under sliding with low loads of cobalt base alloys do not seem to be influenced by the initial grain size. The study showed that a low stacking fault energy and discrete planar sliding under plastic deformation are required to attain a steady increase in strength towards the surface without generating any defect structures which might increase the wear rate. With this knowledge, one may think of other alloys for all-metal prosthesis, e.g. high-nitrogen stainless steels which provide a low stacking fault energy and comparable wear rates in self-mating contact [53]. The future work should envisage the distribution of the interfacial medium and the environment within the mechanically mixed nanocrystalline zone. In addition the fuzzy microstructure between the “mixed” zone and the dynamically recrystallized should be investigated to more detail. Last but not least it is important to notice, that even if the supposed grain boundary hardening is verified by static nanoindentation techniques it is still not verified for the dynamics of such a contact situation. Dynamic recrystallization is characterized by grain rotation, which normally suppresses the materials ability to strain harden. As this is published for rotating subgrains of high SFE materials it is questionable, whether this is also true or false for rotating rhomboid cells of stacking faults or -martensite needles.
Acknowledgements The authors would like to thank gb Implantattechnik GmbH, Essen, Germany for providing wrought cobalt base alloy. In addition the authors are indebted to Prof. Dr. W. Dudzinski from the Wroclaw University of Technology in Poland and B. Gleising from the University of DuisburgEssen for their extensive collaboration with TEM preparation and evaluation.
896
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
References [1] H. Seifert, Vergleichende experimentelle Untersuchungen zum Steifigkeits-, Verformungs- und D¨ampfungsverhalten unterschiedlicher H¨uftgelenktotalendoprothesen unter dynamischer Belastung, Orthop¨adie 26 (1997) 166–180. [2] T.M. Wright, S.B. Goodman (Eds.), Implant Wear in total joint replacement: clinical and biologic issues, Material and Design Considerations. A symposium held in October 2000; Supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases, ISBN number 0-89203-261-8. [3] P. Fehsenfeld, Hochaufl¨osende Verschleißdiagnostik f¨ur die Endoprothesenentwicklung, Nachrichten FZK 32 (1–2) (2000) 91– 96. [4] H.G. Willert, M. Semlitsch, Reactions of the articular capsule to wear products of artificial wear prostheses, J. Biomed. Mat. Res. 11 (1977) 157–164. [5] G. T¨ager, E. Euler, W. Plitz, Changes in the surface of McKee-Farrar arthroplasties, Orthop¨ade 26 (1997) 142–151. [6] G. T¨ager, Analyse von Metall/Metall-Gleitpaarungen bei H¨uftgelenken der ersten Generation im Hinblick auf Grundlagen f¨ur eine neue Anwendung, M.D. Thesis, Ludwig-MaximiliansUniversitaet M¨unchen, Munich, Germany, 1998. [7] C. Sprecher, Charakterisierung der Ablagerungen an Metall-MetallPaarungen von H¨uftendoprothesen, Diploma Thesis, AO Research Davos, Switzerland, 2002. [8] M.A. Wimmer, J. Loos, M. Heitkemper, A. Fischer, The acting wear mechanisms on metal-on-metal hip joint bearings – in-vitro results, Wear 250 (2001) 129–139. [9] M.A. Wimmer, C. Sprecher, R. Hauert, G. T¨ager, A. Fischer, Tribochemical reactions on metal-on-metal hip joint bearings – a comparison between in-vitro and in-vivo results, Wear 255 (2003) 1007–1014. [10] R. B¨uscher, B. Gleising, W. Dudzinski, A. Fischer, Durchstrahlungselektronenmikroskopische Untersuchungen an explantierten MetallMetall H¨uftgelenken, Praktische Metallographie, in print. [11] R. B¨uscher, A. Fischer, Metallurgical Aspects of Sliding Wear of fcc Materials for Medical Applications, Matwer 34 (2003) 10/11, 966–975. [12] M.E. Sikorski, The adhesion of metals and factors that influence it, Wear 7 (1964) 144–162. [13] R. B¨uscher, G. T¨ager, W. Dudzinski, B. Gleising, M.A. Wimmer, A. Fischer, Subsurface Microstructure of Metal-on-Metal Hip Joints and its Relationship to Wear Particle Generation, J. Biomed. Mater. Res. Appl. Biomat. 72B (2005) 206–214. [14] P.J. Firkins, J.L. Tipper, M.R. Saadatzadeh, E. Ingham, M.H. Stone, R. Farrar, J. Fisher, Quantitative analysis of wear and wear debris from metal-on-metal hip prostheses tested in a physiological hip joint simulator, Biomed. Mater. Eng. 11 (2001) 143–157. [15] P.F. Doorn, P.A. Campbell, J. Worrall, P.D. Benya, H.A. McKellop, Metal wear particle characterization from metal on metal total hip replacements: TEM study of periprosthetic tissues and isolated particles, J. Biomed. Mater. Res. 42 (1998) 103–111. [16] A.S. Shanbag, J.J. Jacobs, J. Black, Macrophage/particle interactions: Effect of size, composition and surface area, J. Biomed. Mater. Res. 28 (1994) 81–90. [17] C.P. Case, V.G. Langkammer, C. James, M.R. Palmer, A.J. Kemp, P.F. Heap, L. Salomon, Widespread dissemination of metal debris from implants, J. Bone Joint Surg. 76 (1994) 701–712. [18] J.D. Bobyn, Tribology of Metal/Metal Articulations, AAOS Annual Meeting, Anaheim 1999. [19] I. Bos, R. Johannisson, TEM Darstellung der Abriebpartikel von Gelenkendoprothesen und von ultrastrukturellen Zellver¨anderungen, Biomed. Tech. 48 (2003) 20–26. [20] Z. Lu, Y.B. Xu, Z.Q. Hu, Low cycle fatigue behaviour of a DZX40M directionally solidified co-base superalloy, Acta Metall. Sin. A33 (1997) 1239–1245.
[21] Z. Lu, Y.B. Xu, Z.Q. Hu, Evolution of dislocation structure induced by cyclic deformation in a directionally solidified cobalt base superalloy, Mater. Sci. Technol. 15 (1999) 165–168. [22] P. Huang, H.F. Lopez, Effects of grain size on development of athermal and strain induced -martensite in CoCrMo implant alloy, Mater. Sci. Techn. 15 (1999) 157–164. [23] E.O. Hall, The deformation of mild steel: III discussion of results, Proc. Phys. Soc. B64 (1951) 747–753. [24] N.J. Petch, The cleavage strength of polycrystals, J. Iron Steel Inst. 174 (1953) 25–28. [25] M. Umemoto, Nanocrystallization of steels by severe plastic deformation, Mater. Trans. 44 (10) (2003) 1900–1911. [26] E. Arzt, Size effects in materials due to microstructural and dimensional constraints: a comparative review, Acta Mater. 46 (16) (1998) 5611–5626. [27] T.G. Nieh, J. Wadsworth, Hall-Petch relation in nanocrystalline solids, Scripta Metall. Mater. 25 (4) (1991) 955–958. [28] S. Takeuchi, The mechanism of the inverse Hall-Petch relation of nanocrystals, Scripta Mater. 44 (8–9) (2001) 1483–1487. [29] E. Hellstern, H.J. Fecht, Z. Fu, W.L. Johnson, Structural and thermodynamic properties of heavily mechanically deformed Ru and AlRu, J. Appl. Phys. 65 (1) (1988) 305–310. [30] H. Van Swygenhoven, A. Caro, D. Farkas, Grain boundary structure and its influence on plastic deformation of polycrystalline fcc metals at the nanoscale: a molecular dynamics study, Scripta Mater. 44 (8–9) (2001) 1513–1516. [31] V. Koinkar, Micro/nanotribology and ist application to magnetic media, heads and MEMS, Ph.D. Dissertation, The Ohio State University, 1996. [32] D.A. Rigney, in: Proc. Mat. Sol. ‘97 on Wear of Eng. Mat., 15–18 September, 1997, Indianapolis, Indiana, USA, Microstructural evolution during sliding (1997) 3–12. [33] J.H. Dautzenberg, The role of dynamic recrystallization in dry sliding wear, Wear 60 (1980) 401–411. [34] P. Heilmann, D.A. Rigney, Running-in process affecting friction and wear, in: D. Dowson, et al. (Eds.), The Running-In Process in Tribology, Butterworth, Guilford, England, 1982, p. 25. [35] W.J. Saleski, R.M. Fisher, R.O. Ritchie, G. Thomas, The Nature and origin of sliding wear debris from steels in: K.C., Ludema (ed.), Proc. Conf. “Wear of Materials 83”, 11,14.4.1983 345 East 47th Street, New York, NY. 10017, USA Reston, VA, USA, ASME, 1983, 434. [36] H. Berns, A. Fischer, Tribological stability of metallic materials at elevated temperatures, Wear 162–164 (1993) 441–449. [37] J.A. Hines, K.S. Vecchio, S. Ahzi, A model for microstructure evolution in adiabatic shear bands, Metall. Mater. Trans. 29A (1998) 191–203. [38] M.A. Meyers, H.R. Pak, Observation of an adiabatic shear band in titanium by high-voltage transmission electron microscopy, Acta Metall. 34 (1986) 2493–2499. [39] J.H. Beatty, L.W. Meyer, M.A. Meyers, S. Nemat-Nasser, Formation of controlled adiabatic shear bands in AISI 4340 high strength steel, in: M.A. Meyers, L.E. Murr, K.P. Staudhammer (Eds.), HighStrain-Rate Phenomena in Materials, Dekker, New York, 1992, pp. 645–656. [40] R.W. Armstrong, C.S. Coffey, W.L. Elban, Adiabatic heating at a dislocation pile-up avalanche, Acta Metall. 30 (1982) 2111–2116. [41] A. Fischer, Einfluss der Temperatur auf das tribologische Verhalten metallischer Werkstoffe, Habilitation-Thesis, Ruhr University Bochum, 1992 s.a. VDI-Fortschr. Ber. VDI Z, Reihe 5, Bd. 378, VDI Verlag D¨usseldorf, Germany, 1994. [42] M.A. Meyers, Y.B. Xu, Q. Xue, M.T. Perez-Prado, T.R. McNelley, Microstructural evolution in adiabatic shear localization in stainless steel, Acta Mat. 51 (2003) 1307–1325. [43] D.A. Rigney, W.A. Glaeser, The significance of near surface microstructure in the wear process, Wear 46 (1978) 41–46. [44] A.P. Sutton, Interfaces in crystalline materials, Clarendon Press, Oxford, 1995.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897 [45] I. Tikhovski, H. Brauer, M. M¨olders, M. Wiemann, D. Bingmann, A. Fischer, Fatigue Behaviour and in-vitro Biocompatibility of the Nifree austenitic high-nitrogen steel X13CrMnMoN18-14-3, in: G.L. Winters, M.J. Nutt (Eds.), Stainless Steels for Medical and Surgical Applications, ASTM STP, vol. 1438, ASTM International, West Conshohocken, PA, 2003, pp. 119–136. [46] I. Tikhovski, B. Gleising, S. Weiß, A. Fischer,I. Tikhovski, B. Gleising, S. Weiß, A. Fischer, Mikrostrukturentwicklung des hochstickstofflegierten austenitischen Stahles X13CrMnMoN 18-14-3 unter zyklischer Beanspruchung. in: A. Kneissl, F. Jeglitsch (Eds.), Fortschritte in der Metallographie. Prakt. Met. Sonderband 34, Werkstoffinformationsgesellschaft, Frankfurt, Germany, 2003, pp. 43–50. [47] H.W. Zhang, Z.K. Hei, G. Liu, J. Lu, K. Lu, Formation of nanostructured surface layer on AISI 304 stainless steel by means of surface mechanical attrition treatment, Acta Mater. 51 (2003) 1871–1881. [48] P. Heilmann, J. Don, T.C. Sun, W.A. Glaeser, D.A. Rigney, Characterization of wear surfaces and wear debris generated by sliding, Wear 91 (1983) 171–190. [49] L.H. Chen, D.A. Rigney, Transfer during unlubricated sliding wear of selected metal systems, Wear 105 (1985) 47–61.
897
[50] D.A. Rigney, J.E. Hammerberg, Unlubricated sliding behaviour of metals, MRS Bull. 23 (6) (1998) 32–36. [51] X.-Y. Fu, D.A. Rigney, M.L. Falk, Sliding and deformation of metallic glass: experiments and MD simulations, J. Non-Crystall. Solids 317 (2003) 206–214. [52] D.A. Rigney, X.Y. Fu, J.E. Hammerberg, B.L. Holian, M.L. Falk, Examples of structural evolution during sliding and shear of ductile materials, Scripta Mater. 49 (10) (2003) 977–983. [53] S. Koch, R. B¨uscher, I. Tikhovskiy, H. Brauer, A. Runiewicz, W. Dudzinski, A. Fischer, Mechanical, chemical and tribological properties of the nickel-free high-nitrogen steel X13CrMnMoN18-14-3 (1.4452), Matwer 33 (2002) 705–715. [54] A. Unsworth, Tribology of human and artificial joints, Proc. Inst. Mech. Engrs. 205 (1991) 163–172. [55] W.A. Hodge, R.S. Fijan, K.L. Carlson, R.G. Burgess, W.H. Harris, R.W. Mann, Contact pressures in the human hip joint in vivo, Proc. Natl. Acad. Sci. 83 (1986) 2879–2883. [56] DIN ISO 5832-12: Implants for surgery – metallic materials – part 12: wrought cobalt-chromium-molybdenum alloy, Beuth, Berlin 1999.
The pathways of dynamic recrystallization in all-metal hip joints夽 R. B¨uscher ∗ , A. Fischer Department of Materials Science II, University of Duisburg-Essen, Lotharstr. 1, 47057 Duisburg, Germany Received 31 August 2004; received in revised form 12 January 2005; accepted 1 February 2005 Available online 10 May 2005
Abstract In the present study metal-on-metal (MoM) McKee-Farrar prostheses of the first generation were investigated by means of transmission electron microscopy (TEM). It was possible to observe the worn regions using a novel taper section preparation technique without producing any artefacts. Thereby, it could be shown that the wear in vivo leads to a reduction in grain size by a factor of up to 20,000. This is achieved by recrystallization via two pathways which act simultaneously within the subsurface regions. One is dominated by the metallurgical characteristics of the material and follows the gradient of the friction induced shear strains from the bulk towards the surface. The other one is merely acting within the tribological contact volume directly at the surface. Both mechanisms lead to a significant increase in strength. © 2005 Elsevier B.V. All rights reserved. Keywords: Recrystallization; Nanocrystals; Wear; Artificial hip joints
1. Introduction With approximately 750,000 surgeries per year the total or surface replacement of the natural hip joint is one of the most common surgeries in orthopaedics [1]. Artificial hip joints have been used since 1938 and are nowadays implanted with a high chance of success, e.g. 97% in 2000 [2]. However, the postoperative rate of revisions is still high with 50% in 1996 [3]. Although many different effects like individual medical predisposition or life style and to a much lesser extent clinical mistakes influence the life time of artificial hip joints, the most dominant reason for early failure is still induced by cellular foreign body reactions resulting from the emission of wear particles [4]. Therefore, minimizing wear in the articulating contact areas is still the major goal to attain orthopaedic implants with a longer lifetime. The tribological demands to achieve clinical lifetimes beyond 15 years are currently fulfilled by four material combinations: metal-on-polymer, ceramic-on-polymer, ceramic夽 To be presented at, Wear of Materials Conference 2005, San Diego, CA, USA. ∗ Corresponding author. Tel.: +49 203 379 1266; fax: +49 203 379 4374. E-mail address: [email protected] (R. B¨uscher).
0043-1648/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2005.02.036
on-ceramic and metal-on-metal. The choice of material does not only depend on the friction and wear properties but is also strongly influenced by the expected lifetime and activity – thus the age – of the patient, the experience of the surgeon and finally also the financial and legal situation of the clinic. In the US, as the world’s largest market with about 270,000 total hip replacements each year [2], metal-on-polymer couples are still referred to as the “golden standard”. However, a comparison of the wear rates of the different pairings shows that the classic Charnley prosthesis (metal-on-polymer) generally leads to higher clinical wear rates (<500 m/a) when compared to hard-hard couples such as ceramic-on-ceramic and metal-on-metal joints. The latter are believed to attain very low wear rates of 1–5 m/a after an intense running-in wear in the order of 35 m/a. The fact that some of the joints from the first generation of MoM hip joints (e.g. McKeeFarrar prostheses) lasted for more than 20 years, led to a promising renaissance of all-metal pairings. Nevertheless, the metallurgical mechanisms, which contribute to high runin wear as well as to low steady state wear rates are yet to be understood. Therefore, the present study will focus on the metallurgical investigation of the microstructural changes, which lead to the acting in vivo wear mechanisms of MoM hip prostheses.
888
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
2. Materials and methods
Table 2 Clinical data of the retrieved and studied McKee-Farrar couples (from [5])
Since the number of retrievals is always limited and – as to the nature of human life style – in vivo wear process takes place under distinctly unidentified parameters (starting conditions, loading sequences, duration, and joint motion characteristics, etc.) specimens from a conventional disc-on-pin (DoP) wear test were included for comparison in order to investigate the impact of mechanically dominated wear mechanisms on the subsurface microstructure. Thus all microstructural investigations were performed on laboratory wear specimens and on retrievals.
Element
Couple A
Couple B
Couple C
Alloy trade name Manufacturer
Coballoy Dow, UK
Head diameter Age of patient at revision Duration in vivo
35 mm 80 a
Vinertia Howmedica, USA 40 mm 69.4 a
Vinertia Howmedica, USA 35 mm 85.3 a
7.5 a
17.8 a
19 a
2.1. Laboratory wear specimens Details about the testing equipment and conditions can be found in Table 1. A fixed cylindrical pin with a diameter of 6 mm and a spherical tip (radius: 16 mm) runs against a rotating disc with a diameter of 30 mm. Both are emerged in distilled water. The radius of the as generated wear track was 9 mm and the frequency of testing was 1.77 Hz. A normal load of 5 N was applied resulting in a theoretical maximum Hertzian contact pressure of 370 MPa. Prior to testing, the contacting surfaces were mechanically polished to a Ra value of 0.02 m. The total sliding distance was 10.4 km corresponding to 30 h testing period. This would be roughly 10 weeks in vivo. It should be made clear that the focus of this additional laboratory study was not to simulate body conditions but to solely investigate and understand microstructural changes under tribologically induced shear stresses. This should be done under the action of mechanically dominated wear mechanisms, which are similar to those in vivo. Certainly the influence of chemically dominated wear mechanisms is not regarded. But with respect to the subsurface shear stresses they might be neglected in a first approach. Incorporating the chemistry will be a substantial part of the authors’ future work after the metallurgy has been fully understood. 2.2. Retrievals (in vivo) Three metal-on-metal (MoM) couplings were chosen from a set of 42 retrieved McKee-Farrar prostheses on the basis of Table 1 In vivo conditions vs. laboratory test conditions
typical wear appearance. The 42 retrievals were in situ for 13.6 years at average (range 1, 3 . . . 22) and none of them was removed for the reason of excessive wear. At the time of removal the prosthesis were carefully rinsed to remove blood and subsequently sterilized and packed. Caution was taken not to touch or damage the bearing surfaces. The entire collection of 42 retrievals has been obtained from a single surgeon and intensively studied by the authors with respect to wear appearances, wear mechanisms and their sequence of acting [5–9]. For this contribution three pairs were solely selected on the basis of macroscopic wear appearances. There was no effort undertaken to control for additional parameters like e.g. manufacturer or time in situ. All three implants showed wear marks in the form of fine scratches or grooves as well as tribochemical reaction layers [9], but no major damage of the articulating surfaces was observed by the unaided eye. Details about brand name, manufacturer and clinical data of those three prostheses are summarized in Table 2. 2.3. Materials The laboratory testing material was a forged low carbon CoCrMo-alloy (Endocast SL, gb Implantattechnologie, Essen, Germany), with a chemical composition similar to that of the cast McKee-Farrar prostheses (Table 3) The major differences lie within the grain size of 600 m for the cast endoprotheses and of 37 m for the forged alloy. In addition the latter does not contain as many carbides. However, neither Endocast SL nor the McKee-Farrar alloys belong to the Table 3 Chemical composition (in wt.%) of the studied low carbon wrought alloy CoCr29Mo6 (Endocast SL) relative to the McKee-Farrar alloys (Coballoy, Vinertia) and ISO 5832/12 [56]
Parameter
Cup-on-head
Disc-on-pin
Element
Endocast SL
Coballoy
Vinertia
ISO 5832/12
Contact geometry
27.5 5.37 0.008 0.17 <0.002 0.26 0.08 0.01 0.42 0.53
26–30 5–7 <0.35 <0.25
<1 2.9
<1 2.3
<0.75 <1
<1
<1
Synovial fluid
Cr Mo C N S Fe Ni W Si Mn Al Co
27.4 5.4
Medium
Flat versus convex (radius: 16 mm) 5N 370 MPa 0.1 m/s (1.77 Hz) 22 ◦ C Unidirectional 30 h (2.8 km sliding distance) Distilled water
26.4 4.5
Normal force Hertzian pressure Relative velocity Temperature Sliding motion Duration of loading
Concave versus convex <2000 Na <20 MPab <0.2 m/sa (1 Hz) 37 ◦ C Multidirectional 7.5 a
<1 <1
<1 Balance
<1 Balance
a b
Unsworth [54]. Hodge et al. [55].
Balance
Balance
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
so-called high carbon (HC) cast or forged alloys, which had been investigated earlier [8]. Certainly, the microstructure, morphology and volume fraction of carbides strongly influence the wear rate. Thus, this work will not compare wear rates and try to generalize the results for the various alloy types but will address the basic microstructural changes of low carbon cobalt base alloys under shear strains induced by sliding wear. 2.4. Metallography, microscopy, and hardness testing
889
specimens were further thinned from both sides by means of dimple grinding (Model 656, Gatan GmbH, Munich, Germany) and ion milling (PIPS 691, Gatan, Munich, Germany) down to 100 nm. TEM investigations were performed with an accelerating voltage of 120 keV. In a recent study with worn high-nitrogen steel samples it has been demonstrated that this technique does not introduce artefacts [10]. Since the physiological gait cycle creates a multidirectional movement on the surface of the hip joint, the orientation of the in vivo taper section can not be chosen according to a certain direction of relative motion. However, the DoP wear test is unidirectional and, therefore, this allows for the defined orientation of the samples during preparation. But, due to the fact that the wear track is narrow and circular the chosen TEM preparation technique allows only a preparation of a transversal taper section. Thus, the figures will show the microstructure underneath the wear track perpendicular to the sliding direction. In order to evaluate the strain induced hardening of the subsurface microhardness measurements were carried out by means of a Vickers indenter and a load of 10 p (=98 mN, HV0.01). The hardness gradient was measured at 40 positions within a distance of up to 40 m below the surface.
In order to gain information about the hypothetical changes of the microstructure underneath the articulating surfaces, in vivo and DoP samples were investigated in taper sections (=90◦ cross section) using a light microscope (LiMi) as well as a transmission electron microscope (EM 400, Phillips, Eindhoven, The Netherlands). For LiMi taper sections from heads and cups as well as from disc’s and pins, samples were prepared by wet cutting, grinding, polishing and finally chemical etching for 45 s in 50 ml H2 O + 50 ml HCl + 4 g K2 S2 O5 . For TEM a conventional preparation technique was employed using two parallel cuts of the worn areas with a thickness of 500 m, which were glued together with a twocomponent adhesive. Fig. 1 shows a medial and parallel cut of the prostheses with the sample regions indicated with 1 (cup sample) and 2 (head sample). To minimize the gap between the samples, a convex sample of the head was glued together with a concave counterpart from the joint cup. The preparation of the DoP samples was much easier done by glueing together two pieces of one disc. These prepared samples were then fixed in a brass cylinder (diameter: 3 mm) using a slotted pipe (diameter: 2.5 mm) as well as a suitable adhesive (Epoxy G1, Gatan, Munich). A heat treatment at room temperature for 30 min and at 150 ◦ C for 2 h resulted in a sufficient bonding strength of these composite structures. Afterwards it was cut into 400 m thin slices by slow speed wet cutting. After grinding to a thickness of 100 m, the
At a certain magnification (Fig. 2) a distinct degree of plastic deformation can be observed below the worn track in the direction of the tribologically introduced shear stresses. According to the LiMi observations, TEM reveals different appearances of plastic deformation with respect to the distance to the worn surfaces. At a depth of about 30 m from the worn surface stacking faults and -martensite needles prevail (Fig. 3). Electron diffraction patterns of the needles verify -martensite, which has a hexagonal closed packed
Fig. 1. In vivo; cross-sectional overview of a cut McKee-Farrar prosthesis.
Fig. 2. Light microscopic cross-section of the wear track. The arrow indicates the sliding direction of the pin. NC = nanocrystalline zone.
3. Results 3.1. Laboratory wear specimens
890
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
Fig. 3. Coexistence of stacking faults and -martensite needles about 30 m below the worn surface (laboratory sample).
Fig. 5. Nanocrystalline microstructure at the disc’s surface (laboratory test).
(hcp) lattice structure and form strictly on the discrete (1 1 1) sliding planes [11]. These lattice defects react with each other under higher shear strain (= closer to the surface) forming an arrangements of, e.g. rhombic cells (Fig. 4). Between depths of about 10 and 1 m no clear microstructure can be identified. Moreover, the metal appears to be in some state of severe plastic deformation. Nevertheless, closer towards the surface at depth of less than about 1 m again a distinct microstructure of very fine granular crystals in the nm range occur (Fig. 5). The microhardness profile (Fig. 6) shows the cold working of the CoCrMo solid solution by the observed lattice defects. Even though the measured values scatter, the tendency is unequivocal. Starting from the microhardness of the unworn material of 450 HV0.01 the hardness increases at about a depth of 30 m up to 660 HV0.01 at 3 m. Closer to the surface the hardness might even be higher, but this cannot be
measured by such microhardness techniques. The measurements adjacent to the wear track prove that the influence from mechanical polishing can be neglected since no increase in hardness was observed within the same region. Therefore, the changes in hardness should be a result of the wear-induced changes of the subsurface microstructure. The coefficient of friction in the steady state wear regime was µ = 0.3.
Fig. 4. Rhombic cells generated by intersecting -martensite needles about 4 m below the worn surface (laboratory test).
3.2. Retrievals LiMi observation reveals the coarse grain structure of the cast material but no significant changes of the microstructure in the vicinity of the contact area as depicted in Fig. 7. Nevertheless, the results of the TEM investigation show a different picture, which appears similar to that of the laboratory specimens. The microstructure below the articulating surface has changed significantly because of tribological stresses. This was observed on both sliding partners of all investigated pairs. Again in a certain depth from the surface, e.g. twins and martensite needles form on the sliding planes of the CoCrMo solid solution (Fig. 8 ). In other areas stacking faults appear as well. Again these lattice defects react with each other generating triangular or rhombic cells. Closer to the surface the grains become nanocrystalline (Fig. 9). The microhardness profile of this retrieval also shows (Fig. 10) cold working but the measured values scatter much more distinct than for the laboratory specimens. Already within the bulk material the values scatter between 310 and 360 HV0.01. This can be simply attributed to the effects of microsegregation within cast microstructures, which show the characteristic dendritic structure within large grains (Fig. 7). Despite the scatter again the work hardening tendency towards the surface is apparent. Beginning with the average microhardness of the unworn material of about 340 HV0.01 the hardness increases at about a depth of 15–20 m up to 390 HV0.01 at 3 m from the worn surface.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
891
Fig. 6. Depth profile of microhardness HV0.01 (laboratory test).
Due to the distinct scatter there is no use in any attempt to quantify the work hardening effect, but in comparison to the laboratory specimens less volume is affected and less work hardening is noticed in the region below 2 m from the worn surface. In addition the in vivo gradient is flatter. Nevertheless, the work hardening is induced by tribologically introduced shear strains leading to lattice defects and phase transformation starting from about 30 m below the surface.
4. Discussion MoM hip joints are self-mating tribological systems in which sliding wear prevails. Following the literature it has been known for decades that because of their high ductility fcc metals – like CoCrMo alloys – are prone to wear by adhesion [12]. However, neither on the worn surfaces of the joint nor on the surfaces of the laboratory samples any signs
Fig. 7. Light microscopical cross-section of McKee-Farrar hip cup. Arrow points on the articulating surface.
of adhesion were present. As indicated by the authors tribochemical reactions are present on the surfaces and generate layers of denaturated organic material. These layers separate the surfaces and thereby hinder direct metal-on-metal contacts [8,9]. However, this effect is supposedly happening on discrete spots on the surfaces and cannot be responsible for the total absence of adhesion. TEM investigations revealed in addition that the uppermost surface consists of a nanocrystalline microstructure of fcc and strain induced hcp CoCrMo solid solutions [11,13]. These fine grains bring about surface fatigue in a nm-range generating the well known globular and needle like wear debris in the nm range, which also can act as solid lubricant and separates the contacting surfaces. Beside the hypothesised – but not verified – extreme cold working effect of such a fine grain and lattice defect rich zone this change in microstructure has certainly a big impact on the wear mech-
Fig. 8. Formation of rhombic and triangular cells by reaction of lattice defects in a distance of about 5 m from the worn surfaces (McKee-Farrar, retrieval B, cup).
892
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
Fig. 9. Nanocrystalline surface of McKee-Farrar Prostheses (retrieval A, cup).
anisms as has been shown by DoP laboratory tests [11]. Due to the gradient of properties towards the worn surface no instabilities occur which would represent eventual sites for crack initiation. Moreover, as a consequence of the low stacking fault energy there is a total absence of dislocation cells below the worn surfaces. Thus tribologically induced fatigue crack initiation and propagation is shifted towards the top nanocrystalline surface generating just nm small delaminating filaments. Such small filaments are easily crumbled into nanoscopic single particles, as they have been observed in several in vitro and in vivo studies [14–19]. Thus, owing to the low stacking fault energy of CoCrMo alloys together with the bio-tribochemical reactions on the worn surfaces MoM hip joints achieve their low clinical wear rates. A very similar development of lattice defects and metallurgical phase transformation was observed for both tribosystems below the contact areas. However, the depths in which
the different effects take place differ, which mainly can be attributed to different loading conditions. Nevertheless, the thickness of the nanocrystalline zone as well as the grain size within it did not differ significantly between the DoP and the in vivo samples. Fig. 11 shows the distribution of grain sizes. Although the mean grain size below the discs surface (37 nm) is slightly higher than in vivo (32 nm) this should not be overestimated. The thickness of the nanocrystalline layer in both tribological systems is well below 1 m; with being slightly thicker on the laboratory sample (800 nm) than in vivo (550 nm). Now the question remains why materials with such a low stacking fault energy – hindering cross slip and climbing – recrystallize at room temperature from 600 m (cast) and 37 m (forged), respectively to about 30 nm. In addition it is not clear why the final grain size in the worn surfaces of both tribosystems are similar, even though the loading conditions and as a consequence the hardness gradients are different. Thus on the basis of the analyses as well as on existing literature at first the strain hardening and at second the recrystallization should be discussed. Due to the fact, that the metallurgical findings of the in vivo and lab specimens are similar both will be treated together. Different observations are mainly due to the different loading conditions and movement characteristics, respectively. Larger contact pressures and, therefore, supposedly higher friction shear forces as well as the unidirectional relative motion lead to a more distinct straining of the DoP subsurface microstructure compared to the multidirectional character of the in vivo tribosystem with unknown coefficient of friction. 4.1. Microstructure and hardness The strain induced hardening of the subsurface is achieved by different microstructural effects, which act simultaneously. Generally, blocking and subsequent accumulation of dislocations lead to an increase in hardness. Blocking can
Fig. 10. Depth profile of microhardness HV0.01 (McKee-Farrar retrieval A, head).
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
893
Fig. 11. Comparison of the grain size distribution in the nanocrystalline zone of in vivo and laboratory specimens.
be obtained from plastic deformation by either increasing the grain boundary density or creating other lattice defects which also hinder the movement of dislocations. In particular, stacking faults and their intersections are known to distinctly increase the strength by piling up dislocations [20]. Merely at elevated temperatures these intersections may be passed by dislocations to achieve a relaxation of the lattice [21]. However, this form of work hardening is even limited at room temperature since the stacking fault density is increased with increased work hardening which in turn hinders the formation of new stacking faults [22]. In addition to stacking faults a strain induced phase transformation into hcp needles further contribute to this hardening effect. It should be mentioned that the hardening effect of strain induced -martensite is similar to that of stacking faults and differs from that of ␣ -martensite in steels. In addition to these grain boundaries must be regarded as well. The quantitative increase in hardness which can be achieved by grain boundaries was described by Hall and Petch in the early 50’s. Their relationship predicts an increasing hardness with a decreasing grain size (Hardness ∼ 1/diameter0.5 ) [23,24]. This is confirmed for cobalt base alloys when the hardness of fine grained wrought alloys (400–500 HV0.05) is compared to the hardness of cast alloys (300–350 HV0,05) having much coarser grains. However, nanocrystalline metals do not always follow the HallPetch relationship. Umemoto [25] showed that the hardness of nanocrystalline Al, Ti, Ni and Fe is even higher than it was calculated referring to Hall-Petch. In contrast any extreme reduction in grain size might also lead to a softening of the material. This effect, which is often referred to as inverse Hall-Petch relation, is valid for grains which are smaller than dislocation loops and hence, no hardening will take place [26–28]. Fine grains also are able to rotate and slide relative to each other thereby leading to a plastic deformation which is not brought about by the generation and movement of dislocations. Thus hardness might not alter or even decrease [29].
For the cobalt base alloys investigated here lattice defects were still visible within the nanocrystals. Thus it is supposed that the inverse Hall-Petch relation does not apply and the hardness is enhanced towards the surface. This is also supported by the fact that by means of molecular modelling it has been shown that the inverse Hall-Petch relation applies at a grain size below 5–8 nm [30]. Referring to the tribological behaviour of a nanocrystalline surface, both, the increase in hardness as well as the increase in plasticity due to grain rotation and sliding should be beneficial. The latter may allow some small foreign body particles to be entrapped within the surface rather than initiating abrasion, while a high hardness on the other hand reduces the effect of three body abrasion by increasing the resistance against surface fatigue by indentation. 4.2. Recrystallization A comparison of the bulk and surface grain sizes reveals their significant reduction. For the studied McKee-Farrar joints, the mean grain size of the as-cast cobalt base alloy is in the order of 600 m. In contrast, the crystals at the surface are 30 nm in size. Hence, a recrystallization took place which reduced the original grain size by a factor of approximately 20,000 times within a zone thinner than 100 m. For the DoP samples this factor is lower due to finer initial grains of the wrought alloy. Nevertheless, even for the wrought metal the original grain size is reduced by more than 1000 times. Due to the fact that all materials with low stacking fault energies do not recrystallize as easy as e.g. Cu- or Al-alloys by the generation of sub grain boundaries or cross slip and climbing of dislocations the distinct decrease in grain size is worth being investigated. Stacking faults do not have this capability. Thus, CoCrMo alloys require a very high degree of deformation and a very high temperature to be statically recrystallized. Now the question remains, whether this type of recrystallization –
894
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
which maybe is a dynamic one - is either related to thermal or mechanical effects or both. The characteristics of the tribological system in vivo might help to answer this question. Due to the convex (head) – concave (cup) pairing of the artificial hip joint, the conformity of surfaces is very high which generally results in large contact areas, low local contact pressures and finally relatively low contact temperatures. In addition, the low carbon alloys provide fairly smooth surfaces, which reduce the probability of high medium or flash temperatures. Therefore, the contact temperatures should not be in a range above 1100 ◦ C where a thermally induced recrystallization might occur. Even with HC-CoCrMo alloys, where the contact area may be much smaller due to protruding carbides, a mean temperature increase of only T = 60 K was calculated using an approach by Kuhlmann-Wilsdorf [8]. Thus, mainly mechanically dominated effects must lead to the significant reduction in grain size towards the surface. All MoM joints predominantly undergo mixed lubrication. Hence, the metallic contact between metallic asperities cannot be ruled out. As it has been shown by Koinkar [31], even very low contact pressures can lead to plastic deformations in ductile fcc metals, i.e. that the “shakedown criteria” which predicts only elastic deformation for small pressures may not be valid on a micro- or nano-scale [32]. Thus, the plastic deformation, which is introduced into the subsurface of hip and cup or disc and pin, is responsible for the recrystallization. Such deformed zones are generated by the Hertzian contact stress field directly under the worn surfaces which brings about a constant sequence of static and cyclic strains as described by several authors for materials with high stacking fault energies like Cu, Al, and plain steel [33–35] or at high temperatures [36]. Depending on strain, strain rate and temperature these shear stresses bring about recrystallization via cross slip and climbing of dislocations generating subgrains, which immediately start to rotate or slide against each other. Due to the fact, that this new formation of crystals take space during the deformation process and not afterwards by additional heating this is called dynamic recrystallization. These effects have been also been found with other materials during machining or high velocity impact leading to so called shear bands, in which one finds a nanocrystalline microstructure [37–39]. Now even though the ambient temperature might not be high, the local temperatures within these shear bands during machining are significantly increased [40], which facilitates non-discrete dislocation movement. Under tribological stresses similar recrystallization processes appear even though neither the strain rates nor the temperatures are that high [41]. Thus at least again for Cu- and Al-alloys the observed recrystallization can be explained by the subsurface formation of cell walls, which are known from fatigue loading and are generated by cross slip and climbing as well. If well-defined dislocation cells are formed recrystallization is probable under further plastic deformation, since a high defect density in the cell walls may allow cell rotation and the formation of new grain boundaries [42,43]. The rotation becomes possible even without any further incorporation of
small and wide angle grain boundaries because the diffusion rate of defects and atoms at the cell walls is increased by a factor of 107 –108 when compared to the bulk material [44]. Still CoCrMo alloys have a very low stacking fault energy (SFE) and the ambient and frictional temperatures are fairly low. Thus the mentioned mechanisms of dynamic recrystallization are not valid here. Nevertheless some kind of recrystallization must occur to gain 50 nm grains from 600 m ones. From wear and fatigue laboratory investigations on low SFE materials it is known that cyclic shear stresses bring about localized plastic deformation along discrete sliding planes which generates dislocations, twins, stacking faults and subsequently -martensite [45,46,11]. Under the in vivo tribological stresses the subsurface shear strains activate several discrete sliding planes. On these sliding planes, the generated stacking faults and -martensite needles, which neither cross slip nor climb, react with each other and form the depicted triangular or rhombic cells with a size ranging from 100 to 500 nm (Fig. 8). Certainly these cells block any further sliding of the incorporated lattice defects. Now, due to the steady wear rate any subsurface volume element approaches the worn surfaces and, therefore, undergoes higher cyclic shear stresses. By further increasing the degree of plastic deformation towards the surface, these cells cannot move. Thus, they are sheared which leads to a relative displacement “s” of such lattice defects as highlighted in Fig. 12. If by further shear strains the displacement exceeds a certain value one may speak of a new grain boundary resulting in a highly textured microstructure with smaller grain sizes. This recrystallization mechanism has been theoretically described earlier by Zhang et al. [47]. With the retrieved McKee-Farrar prostheses, this has been predominantly observed in a depth of less than 5 m from the worn surface (Fig. 8), while for the
Fig. 12. Formation of rhombic substructures: rhombic cell C is shifted by the displacement s relatively to A and B along the fcc sliding planes. If s exceeds certain values, a planar mismatch between the cell B and C divides the former macroscopic grain into nanoscopic subgrains.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
DoP specimens this happens in a distance up to 10 m. In this area this type of dynamic recrystallization generated by tribologically induced shear stresses from the moving Hertzian contact stress field is completed. Thus the final grain size is governed predominantly by the size of the former cells. Nevertheless closer to the surface a zone with no distinctly contrasted microstructure is visible. This may be attributed to the high density of sheared cells, which cannot become smaller anymore and might, therefore, start to rotate – similar to subgrains – and subsequently destroy the texture. But at this time it is just a hypothesis. In contrast to this transitional region, the size of the nanograins directly at the surface is smaller than the size of the cells and might be of different reason. Again the nm range of grains size implies some kind of recrystallization mechanism with a similar appearances as it has been found in Cu and stainless steels of various SFEs and were described as “mechanically mixed” and “material transfer layers” [32]. As a matter of fact it could have been shown that a transfer of material does takes place in all-metal pairings which is predominantly directed from the softer to the harder metal [48,49]. In the case of the artificial hip joints such a proof is hard to give since the pairings are self-mating, that means the contacting metals have the same hardness and the same chemical composition. But if this is true the nanocrystalline layer is a mixture of base material, interfacial medium and environment. Thus at least the chemistry at the grain boundaries should differ from that of the grains itself. In addition some of the grains could be of oxidic or organic nature and stem from incorporated remains of some tribochemical reactions at the surfaces. But the verification with such high resolutions is difficult and could not be brought about up to now. Nevertheless, this type of material transfer by mechanical mixing on a nm scale is quite likely at such surfaces. Molecular dynamics showed that contacting surfaces of ductile metals behave like viscous fluids under shearing [50–52]. Hence, the contacting surfaces are exposed to a permanent plastic deformation and rupture in the micro- and nano-scale, by which the nanocrystals are formed by rotations of accidentally generated clusters of atoms within contacting crystals. Thus, the mean grain size within this layer depends only on the flow characteristics of the materials within such a contact. According to Hellstern et al. [29] there is a minimum grain size for each metal, for which a further reduction becomes unlikely due to the need of very high local stresses. In the wear test as well as in vivo, no grains markedly smaller than 10 nm have been observed from which one can assume that this may be the range of minimum grain size for cobalt base alloys to be generated by tribologically induced shear strains. Nevertheless, smaller grains may be tribologically formed without being detected in this work. Due to the fact that in both tribosystems the CoCrMo alloys undergo the same strain range starting from the bulk one can understand why the resulting grains sizes are similar, even though the strain gradients are distinctly different. In addition the results of the TEM essentially indicated that
895
two separate but simultaneously acting mechanisms achieve the strain induced recrystallization in all-metal hip joints.
5. Conclusions and outlook By means of transmission electron microscopy the subsurface microstructures of metal-on-metal McKee-Farrar prostheses were investigated. The results showed that cobalt base alloys are able to adjust their microstructure to the applied tribological loads by a gradual decrease of the grain size. The dynamic recrystallization is achieved by two mechanisms which are supposed to act simultaneously. One is acting at the top surface – by rotating clusters of atoms – within a “mechanically mixed” zone and the second is acting within the bulk – by shearing of cells generated by stacking faults and -martensite needles - moving towards the surface with the increasing strain gradient. Since laboratory tests with wrought alloys revealed a similar tribologically induced microstructure, the wear mechanisms during the steady state wear under sliding with low loads of cobalt base alloys do not seem to be influenced by the initial grain size. The study showed that a low stacking fault energy and discrete planar sliding under plastic deformation are required to attain a steady increase in strength towards the surface without generating any defect structures which might increase the wear rate. With this knowledge, one may think of other alloys for all-metal prosthesis, e.g. high-nitrogen stainless steels which provide a low stacking fault energy and comparable wear rates in self-mating contact [53]. The future work should envisage the distribution of the interfacial medium and the environment within the mechanically mixed nanocrystalline zone. In addition the fuzzy microstructure between the “mixed” zone and the dynamically recrystallized should be investigated to more detail. Last but not least it is important to notice, that even if the supposed grain boundary hardening is verified by static nanoindentation techniques it is still not verified for the dynamics of such a contact situation. Dynamic recrystallization is characterized by grain rotation, which normally suppresses the materials ability to strain harden. As this is published for rotating subgrains of high SFE materials it is questionable, whether this is also true or false for rotating rhomboid cells of stacking faults or -martensite needles.
Acknowledgements The authors would like to thank gb Implantattechnik GmbH, Essen, Germany for providing wrought cobalt base alloy. In addition the authors are indebted to Prof. Dr. W. Dudzinski from the Wroclaw University of Technology in Poland and B. Gleising from the University of DuisburgEssen for their extensive collaboration with TEM preparation and evaluation.
896
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897
References [1] H. Seifert, Vergleichende experimentelle Untersuchungen zum Steifigkeits-, Verformungs- und D¨ampfungsverhalten unterschiedlicher H¨uftgelenktotalendoprothesen unter dynamischer Belastung, Orthop¨adie 26 (1997) 166–180. [2] T.M. Wright, S.B. Goodman (Eds.), Implant Wear in total joint replacement: clinical and biologic issues, Material and Design Considerations. A symposium held in October 2000; Supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases, ISBN number 0-89203-261-8. [3] P. Fehsenfeld, Hochaufl¨osende Verschleißdiagnostik f¨ur die Endoprothesenentwicklung, Nachrichten FZK 32 (1–2) (2000) 91– 96. [4] H.G. Willert, M. Semlitsch, Reactions of the articular capsule to wear products of artificial wear prostheses, J. Biomed. Mat. Res. 11 (1977) 157–164. [5] G. T¨ager, E. Euler, W. Plitz, Changes in the surface of McKee-Farrar arthroplasties, Orthop¨ade 26 (1997) 142–151. [6] G. T¨ager, Analyse von Metall/Metall-Gleitpaarungen bei H¨uftgelenken der ersten Generation im Hinblick auf Grundlagen f¨ur eine neue Anwendung, M.D. Thesis, Ludwig-MaximiliansUniversitaet M¨unchen, Munich, Germany, 1998. [7] C. Sprecher, Charakterisierung der Ablagerungen an Metall-MetallPaarungen von H¨uftendoprothesen, Diploma Thesis, AO Research Davos, Switzerland, 2002. [8] M.A. Wimmer, J. Loos, M. Heitkemper, A. Fischer, The acting wear mechanisms on metal-on-metal hip joint bearings – in-vitro results, Wear 250 (2001) 129–139. [9] M.A. Wimmer, C. Sprecher, R. Hauert, G. T¨ager, A. Fischer, Tribochemical reactions on metal-on-metal hip joint bearings – a comparison between in-vitro and in-vivo results, Wear 255 (2003) 1007–1014. [10] R. B¨uscher, B. Gleising, W. Dudzinski, A. Fischer, Durchstrahlungselektronenmikroskopische Untersuchungen an explantierten MetallMetall H¨uftgelenken, Praktische Metallographie, in print. [11] R. B¨uscher, A. Fischer, Metallurgical Aspects of Sliding Wear of fcc Materials for Medical Applications, Matwer 34 (2003) 10/11, 966–975. [12] M.E. Sikorski, The adhesion of metals and factors that influence it, Wear 7 (1964) 144–162. [13] R. B¨uscher, G. T¨ager, W. Dudzinski, B. Gleising, M.A. Wimmer, A. Fischer, Subsurface Microstructure of Metal-on-Metal Hip Joints and its Relationship to Wear Particle Generation, J. Biomed. Mater. Res. Appl. Biomat. 72B (2005) 206–214. [14] P.J. Firkins, J.L. Tipper, M.R. Saadatzadeh, E. Ingham, M.H. Stone, R. Farrar, J. Fisher, Quantitative analysis of wear and wear debris from metal-on-metal hip prostheses tested in a physiological hip joint simulator, Biomed. Mater. Eng. 11 (2001) 143–157. [15] P.F. Doorn, P.A. Campbell, J. Worrall, P.D. Benya, H.A. McKellop, Metal wear particle characterization from metal on metal total hip replacements: TEM study of periprosthetic tissues and isolated particles, J. Biomed. Mater. Res. 42 (1998) 103–111. [16] A.S. Shanbag, J.J. Jacobs, J. Black, Macrophage/particle interactions: Effect of size, composition and surface area, J. Biomed. Mater. Res. 28 (1994) 81–90. [17] C.P. Case, V.G. Langkammer, C. James, M.R. Palmer, A.J. Kemp, P.F. Heap, L. Salomon, Widespread dissemination of metal debris from implants, J. Bone Joint Surg. 76 (1994) 701–712. [18] J.D. Bobyn, Tribology of Metal/Metal Articulations, AAOS Annual Meeting, Anaheim 1999. [19] I. Bos, R. Johannisson, TEM Darstellung der Abriebpartikel von Gelenkendoprothesen und von ultrastrukturellen Zellver¨anderungen, Biomed. Tech. 48 (2003) 20–26. [20] Z. Lu, Y.B. Xu, Z.Q. Hu, Low cycle fatigue behaviour of a DZX40M directionally solidified co-base superalloy, Acta Metall. Sin. A33 (1997) 1239–1245.
[21] Z. Lu, Y.B. Xu, Z.Q. Hu, Evolution of dislocation structure induced by cyclic deformation in a directionally solidified cobalt base superalloy, Mater. Sci. Technol. 15 (1999) 165–168. [22] P. Huang, H.F. Lopez, Effects of grain size on development of athermal and strain induced -martensite in CoCrMo implant alloy, Mater. Sci. Techn. 15 (1999) 157–164. [23] E.O. Hall, The deformation of mild steel: III discussion of results, Proc. Phys. Soc. B64 (1951) 747–753. [24] N.J. Petch, The cleavage strength of polycrystals, J. Iron Steel Inst. 174 (1953) 25–28. [25] M. Umemoto, Nanocrystallization of steels by severe plastic deformation, Mater. Trans. 44 (10) (2003) 1900–1911. [26] E. Arzt, Size effects in materials due to microstructural and dimensional constraints: a comparative review, Acta Mater. 46 (16) (1998) 5611–5626. [27] T.G. Nieh, J. Wadsworth, Hall-Petch relation in nanocrystalline solids, Scripta Metall. Mater. 25 (4) (1991) 955–958. [28] S. Takeuchi, The mechanism of the inverse Hall-Petch relation of nanocrystals, Scripta Mater. 44 (8–9) (2001) 1483–1487. [29] E. Hellstern, H.J. Fecht, Z. Fu, W.L. Johnson, Structural and thermodynamic properties of heavily mechanically deformed Ru and AlRu, J. Appl. Phys. 65 (1) (1988) 305–310. [30] H. Van Swygenhoven, A. Caro, D. Farkas, Grain boundary structure and its influence on plastic deformation of polycrystalline fcc metals at the nanoscale: a molecular dynamics study, Scripta Mater. 44 (8–9) (2001) 1513–1516. [31] V. Koinkar, Micro/nanotribology and ist application to magnetic media, heads and MEMS, Ph.D. Dissertation, The Ohio State University, 1996. [32] D.A. Rigney, in: Proc. Mat. Sol. ‘97 on Wear of Eng. Mat., 15–18 September, 1997, Indianapolis, Indiana, USA, Microstructural evolution during sliding (1997) 3–12. [33] J.H. Dautzenberg, The role of dynamic recrystallization in dry sliding wear, Wear 60 (1980) 401–411. [34] P. Heilmann, D.A. Rigney, Running-in process affecting friction and wear, in: D. Dowson, et al. (Eds.), The Running-In Process in Tribology, Butterworth, Guilford, England, 1982, p. 25. [35] W.J. Saleski, R.M. Fisher, R.O. Ritchie, G. Thomas, The Nature and origin of sliding wear debris from steels in: K.C., Ludema (ed.), Proc. Conf. “Wear of Materials 83”, 11,14.4.1983 345 East 47th Street, New York, NY. 10017, USA Reston, VA, USA, ASME, 1983, 434. [36] H. Berns, A. Fischer, Tribological stability of metallic materials at elevated temperatures, Wear 162–164 (1993) 441–449. [37] J.A. Hines, K.S. Vecchio, S. Ahzi, A model for microstructure evolution in adiabatic shear bands, Metall. Mater. Trans. 29A (1998) 191–203. [38] M.A. Meyers, H.R. Pak, Observation of an adiabatic shear band in titanium by high-voltage transmission electron microscopy, Acta Metall. 34 (1986) 2493–2499. [39] J.H. Beatty, L.W. Meyer, M.A. Meyers, S. Nemat-Nasser, Formation of controlled adiabatic shear bands in AISI 4340 high strength steel, in: M.A. Meyers, L.E. Murr, K.P. Staudhammer (Eds.), HighStrain-Rate Phenomena in Materials, Dekker, New York, 1992, pp. 645–656. [40] R.W. Armstrong, C.S. Coffey, W.L. Elban, Adiabatic heating at a dislocation pile-up avalanche, Acta Metall. 30 (1982) 2111–2116. [41] A. Fischer, Einfluss der Temperatur auf das tribologische Verhalten metallischer Werkstoffe, Habilitation-Thesis, Ruhr University Bochum, 1992 s.a. VDI-Fortschr. Ber. VDI Z, Reihe 5, Bd. 378, VDI Verlag D¨usseldorf, Germany, 1994. [42] M.A. Meyers, Y.B. Xu, Q. Xue, M.T. Perez-Prado, T.R. McNelley, Microstructural evolution in adiabatic shear localization in stainless steel, Acta Mat. 51 (2003) 1307–1325. [43] D.A. Rigney, W.A. Glaeser, The significance of near surface microstructure in the wear process, Wear 46 (1978) 41–46. [44] A.P. Sutton, Interfaces in crystalline materials, Clarendon Press, Oxford, 1995.
R. B¨uscher, A. Fischer / Wear 259 (2005) 887–897 [45] I. Tikhovski, H. Brauer, M. M¨olders, M. Wiemann, D. Bingmann, A. Fischer, Fatigue Behaviour and in-vitro Biocompatibility of the Nifree austenitic high-nitrogen steel X13CrMnMoN18-14-3, in: G.L. Winters, M.J. Nutt (Eds.), Stainless Steels for Medical and Surgical Applications, ASTM STP, vol. 1438, ASTM International, West Conshohocken, PA, 2003, pp. 119–136. [46] I. Tikhovski, B. Gleising, S. Weiß, A. Fischer,I. Tikhovski, B. Gleising, S. Weiß, A. Fischer, Mikrostrukturentwicklung des hochstickstofflegierten austenitischen Stahles X13CrMnMoN 18-14-3 unter zyklischer Beanspruchung. in: A. Kneissl, F. Jeglitsch (Eds.), Fortschritte in der Metallographie. Prakt. Met. Sonderband 34, Werkstoffinformationsgesellschaft, Frankfurt, Germany, 2003, pp. 43–50. [47] H.W. Zhang, Z.K. Hei, G. Liu, J. Lu, K. Lu, Formation of nanostructured surface layer on AISI 304 stainless steel by means of surface mechanical attrition treatment, Acta Mater. 51 (2003) 1871–1881. [48] P. Heilmann, J. Don, T.C. Sun, W.A. Glaeser, D.A. Rigney, Characterization of wear surfaces and wear debris generated by sliding, Wear 91 (1983) 171–190. [49] L.H. Chen, D.A. Rigney, Transfer during unlubricated sliding wear of selected metal systems, Wear 105 (1985) 47–61.
897
[50] D.A. Rigney, J.E. Hammerberg, Unlubricated sliding behaviour of metals, MRS Bull. 23 (6) (1998) 32–36. [51] X.-Y. Fu, D.A. Rigney, M.L. Falk, Sliding and deformation of metallic glass: experiments and MD simulations, J. Non-Crystall. Solids 317 (2003) 206–214. [52] D.A. Rigney, X.Y. Fu, J.E. Hammerberg, B.L. Holian, M.L. Falk, Examples of structural evolution during sliding and shear of ductile materials, Scripta Mater. 49 (10) (2003) 977–983. [53] S. Koch, R. B¨uscher, I. Tikhovskiy, H. Brauer, A. Runiewicz, W. Dudzinski, A. Fischer, Mechanical, chemical and tribological properties of the nickel-free high-nitrogen steel X13CrMnMoN18-14-3 (1.4452), Matwer 33 (2002) 705–715. [54] A. Unsworth, Tribology of human and artificial joints, Proc. Inst. Mech. Engrs. 205 (1991) 163–172. [55] W.A. Hodge, R.S. Fijan, K.L. Carlson, R.G. Burgess, W.H. Harris, R.W. Mann, Contact pressures in the human hip joint in vivo, Proc. Natl. Acad. Sci. 83 (1986) 2879–2883. [56] DIN ISO 5832-12: Implants for surgery – metallic materials – part 12: wrought cobalt-chromium-molybdenum alloy, Beuth, Berlin 1999.